Deep Variational Autoencoder (VAE) For 1D unbalanced classification problem for COSCREEN

we’re going to see how How to apply deep convolutional neural networks and auto-encoders for building COSCREEN prediction model.

case(1):

Visualize and Analize data functionality

Handle Categorial Features

There are few categorial features in the data set. Lets view them and create dummy variables.

Encode Categorial Features

Latent space

We can see that SCREEN and Not SCREEN, can be separable at latent space.

classify in Latent space

Any classification method can be used, lets try nearest neighbour * Playing with classification parameter to get best prediction on Validation set

case(2):

Encode Categorial Features

Latent space

We can see that SCREEN and Not SCREEN, can be separable at latent space.

classify in Latent space

Any classification method can be used, lets try nearest neighbour * Playing with classification parameter to get best prediction on Validation set